This paper proposes a fast and reliable two-level islanding detection method (IDM) for grid-connected photovoltaic system (GCPVS)-based microgrid. In the first level of the proposed IDM, the magnitude of the rate of change of output voltage (ROCOV) is computed. If this variable exceeds a predefined threshold, a disturbance is injected into the duty cycle of DC/DC converter after a given time delay to deviate the system operating point away of its maximum power point (MPP) condition. This leads to a substantial active power output and voltage reduction in an islanded mode. Therefore, the ROCOV and the rate of change of active power output (ROCOP) indices, measured in the second stage, pose great negative sets at the same time in islanding states. However, the variation of at least one of these variables is near-zero in non-islanding switching events. The assessment of the presented algorithm has been conducted under extensive islanding and non-islanding scenarios for a case study system with two PV power plants using hardware-in-the-loop (HiL) simulation tests. The provided results remark precise islanding classification with an eminently small non-detection zone (NDZ) within 510 ms. The presented IDM has the advantages of self-standing thresholds determination, no improper effect on the output power quality, and simple and inexpensive structure. Moreover, the fast MPP restoration of the proposed scheme after islanding identification boosts the chance of seamless reconnection and DG autonomous operation in microgrid.

This paper proposes a novel islanding detection method (IDM) for grid-connected photovoltaic systems (GCPVSs) through a disturbance injection in the maximum power point (MPP) tracking algorithm. When an absolute deviation of the output voltage exceeds a threshold, the applied disturbance shifts system operating point from its MPP condition. This leads to a sharp active power output reduction and consequently, a significant voltage drop in islanded mode. The proposed algorithm is defined in a way that the distributed generator (DG) can be restored to MPP after islanding classification. It is thereby effective in microgrid where the power injection at maximum level to cater the critical loads and maintain the stability of the isolated area are pursued. The assessment of the proposed technique has been conducted for a sample network containing two GCPVSs in a real-time platform including actual relays in hardware-in-the-loop (HiL). The provided results under extensive islanding endorse timely and accurately detection with negligible non-detection zone (NDZ) and no false tripping in non-islanding disturbances. The comparative analysis of the presented scheme with a few recent IDMs highlights its overall superiorities, including very small NDZ, fast detection, thresholds self-standing determination, no adverse effect on power quality, and simple and inexpensive integration.

The connection of renewable energy sources (RESs) to the distribution network has been rising at a steady pace over the past decades. The great penetration of RESs such as grid‐connected photovoltaic system brings new technical challenges to the distribution networks such as unintentional islanding. Conceptually, this situation occurs when a portion of the network that has been isolated from the main grid remains energised by the embedded RESs. This unexpected scenario should be thereby identified effectively to avoid frequency and voltage deviations and their hazardous effects. The aim of this paper is to provide a comprehensive review on the recently developed islanding detection methods for grid‐following/grid‐connected photovoltaic system, analyse their existing limitations, and suggest possible future research implementations. In this context, an in‐depth comparison is provided considering the main features used in islanding detection methods such as non‐detection zone, detection time, implementation cost and complexity, and power quality degradation. Finally, the main technical requirements established by the current grid codes are recalled identifying potential multi‐functional approaches to expand the current islanding detection capabilities.